Fake Measurement Detection in Automatic Voltage Control using Support Vector Machine

As a novel data mining method featured with excellent pattern recognition capability, Support vector machine (SVM) is utilized in this paper to detect erroneous remote measurement, and therefore, prevent mis-operation of automatic voltage control (AVC) systems. A SVM nonlinear regression algorithm is first used to predict remote measurement. The remote measurement deviate obviously from the predicted value is fed to another SVM to identify the erroneous remote measurement, and thereafter, prevent mis-operation of AVC due to erroneous remote measurement. Numerical simulation shows effectiveness of the proposed approach.